Spaceborne bistatic polarimetrie SAR for scattering analysis and classification of man-made and natural features

摘要

Polarimetrie Synthetic Aperture RADAR (SAR), is an emerging technology in the field of remote sensing today which has several applications in a number of study areas. It has a unique feature as it combines the enhanced resolution that is available through the use of SAR with the wave properties of Polarised electromagnetic energy caused due to scattering of radiation as it interacts with different features on earth's surface. A polarimetric backscatter signature is a three-dimensional saddle plot which is created when ellipticity, orientation angle and amplitude of the backscattered polarised wave is graphed. Since different features interact differently with the polarised wave, each man-made and natural feature causes a unique scattering of the polarised wave which is studied in the backscatter signature received. Based on different scattering mechanisms, a number of decomposition models were proposed like Freeman's decomposition model, Sinclair decomposition model, Pauli's decomposition, Yamaguchi's decomposition model and many more. Based on the relative scattering of the polarised wave, the identification and classification of various feature classes was possible. In this study it was aimed to study the different polarimetric decomposition models and the man-made and natural features detected in each, based on the relative scattering mechanisms and their relative classification using machine learning and object oriented approaches with improved classification accuracy.